Atmospheric Chemistry and Physics Absorbing aerosol in the troposphere of the Western Arctic during

In the spring of 2008 NASA and NOAA funded the ARCTAS and ARCPAC field campaigns as contribu- tions to POLARCAT, a core IPY activity. During the cam- paigns the NASA DC-8, P-3B and NOAA WP-3D aircraft conducted over 160 h of in-situ sampling between 0.1 and 12 km throughout the Western Arctic north of 55 N (i.e. Alaska to Greenland). All aircraft were equipped with mul- tiple wavelength measurements of aerosol optics, trace gas and aerosol chemistry measurements, as well as direct mea- surements of the aerosol size distributions and black car- bon mass. Late April of 2008 proved to be exceptional in terms of Asian biomass burning emissions transported to the Western Arctic. Though these smoke plumes account for only 11-14 % of the samples within the Western Arc- tic domain, they account for 42-47 % of the total burden of black carbon. Dust was also commonly observed but only contributes to 4-12 % and 3-8 % of total light absorption at 470 and 530 nm wavelengths above 6 km. Below 6 km, light absorption by carbonaceous aerosol derived from ur- ban/industrial and biomass burning emissions account for 97-99 % of total light absorption by aerosol. Stratifying the data to reduce the influence of dust allows us to determine mass absorption efficiencies for black carbon of 11.2 ±0.8, 9.5±0.6 and 7.4±0.7 m 2 g 1 at 470, 530 and 660 nm wave-


Introduction
The cryosphere, perhaps more than any other domain, is experiencing a profound, perhaps millennia-long (Solomon et al., 2009), transformation that is in part a result of anthropogenic induced climate change (ACIA, 2004;IPCC, 2007).Yet in many respects, the Arctic atmosphere remains critically undersampled.Ground based measurements of Arctic aerosol have been made since the early 1980's (Quinn et Published by Copernicus Publications on behalf of the European Geosciences Union.al., 2007) but coverage is sparse and limited to the surface boundary layer.Ground-based remote sensing using sunphotometers (Bokoye et al., 2002;Tomasi et al., 2007) or lidar (Bourdages et al., 2009;Ishii et al., 1999;Leaitch et al., 1989), also have sparse coverage and retrievals complicated by seasonally persistent cloud cover.Satellite retrievals at high latitudes are also complicated by cloud cover and a typically bright target surface (i.e.snow and ice).Although airborne campaigns have provided in-situ measurements throughout the depth of the Arctic troposphere (Clarke et al., 1984;Rogers et al., 2001;Scheuer et al., 2003;Yamanouchi et al., 2005) their temporal and spatial coverage limits the extent to which they can be used to constrain stateof-the-art climate model simulations (e.g.Koch et al., 2009).
Arctic Haze is a winter/spring accumulation of aerosol and trace gases in the Arctic atmosphere thought to be largely of anthropogenic origin (Shaw, 1995).In-situ snow sampling during the International Polar Year (IPY) confirm that urban industrial emissions of black carbon are common in the Arctic snowpack and that biomass burning aerosol is a dominant signal in Spring (Hegg et al., 2010).Potential increases in wildfire frequency and intensity in a warmer climate (Stocks et al., 1998;Westerling et al., 2006) have led to renewed interest in the role of fire in the Earth's climate system (Fromm et al., 2010).However, reliable estimates of plume injections heights, required for accurate simulation of their long-range transport and radiative effects, have only recently become available on global scales (Guan et al., 2010).
The radiative effects of the absorbing components of haze are enhanced over a reflective surface and, unlike snowfree regions, the radiative effects continue by reducing snow albedo after their removal to the snow pack (Clarke and Noone, 1985).Reductions in snow and ice albedo is of particular concern because of its positive feedback on melting of seasonal snow cover at mid-latitudes (Flanner et al., 2009), permanent snow and ice at high altitude (Menon et al., 2010;Xu et al., 2009) and in the polar regions (Hansen and Nazarenko, 2004).
In the spring of 2008 NASA-funded the Arctic Research of the Composition of the Troposphere from Aircraft and Satellites (ARCTAS) field campaign (Jacob et al., 2010), in conjunction with the NOAA-funded Aerosol, Radiation, and Cloud Processes affecting Arctic Climate (ARCPAC) (Brock et al., 2010).These experiments were US contributions to POLARCAT, a core International Polar Year (IPY) activity.
Here we summarize measurements of absorbing aerosol in the troposphere of the Western Arctic including black carbon (BC), brown carbon (BrC) and mineral dust.In this paper we use gas-phase tracers and aerosol composition to discriminate absorbing aerosol of urban/industrial origin from absorbing aerosol generated from biomass burning.This method of stratification is then applied to simultaneous measurements of carbon monoxide, accumulation mode black carbon mass and aerosol optical properties.This enables the determination of mass absorption efficiencies (MAE) for these aerosol types over the Western Arctic and leads to a discussion of their relative contributions to total light absorption by aerosol sampled in the troposphere over the Western Arctic in the Spring of 2008.

Instrumentation
Three single particle soot absorption photometers (SP2) (Baumgardner et al., 2004;Stephens et al., 2003;Subramanian et al., 2010) were deployed on board NASA and NOAA research aircraft during the ARCTAS/ARCPAC airborne field campaigns.The DC-8 instrument (Moteki and Kondo, 2007;Moteki et al., 2007) was operated by a team from the University of Tokyo (Kondo et al.), the P-3B instrument was operated by a team from the University of Hawaii (Clarke et al.) and the NOAA WP-3D instrument (Schwarz et al., 2006) was operated by a team from NOAA's Aeronomy Lab (Fahey et al.).Teams from the University of Jimenez et al.), the University of Hawaii (P-3B, Howell) and NOAA's ESRL CSD (WP-3D, Middlebrook et al.) measured submicrometer non-refractory aerosol chemistry (NH + 4 , NO − 3 , SO 2− 4 , NR Cl − , and organics) using high-resolution time-of-flight aerosol mass spectrometry (HR-ToF-AMS) (Canagaratna et al., 2007;DeCarlo et al., 2006).Carbon monoxide is measured on board the DC-8, P-3B and WP-3D by teams from NASA Langley (Diskin), NASA Ames (Podolske) and NOAA ESRL CSD (Holloway) research centers.Gas-phase measurements of acetonitrile were measured on board the DC-8 by the University of Innsbruck (Wisthaler et al.) (Lindinger et al., 1998;Sprung et al., 2001) and aboard the WP-3D by a team from NOAA's ESRL CSD (de Gouw et al.) (de Gouw and Warneke, 2007).Acetonitrile was not measured on board the NASA P-3B.
Total and submicrometer aerosol light scattering were measured onboard the DC-8 by the NASA Langley Research Center (Anderson et al.) and onboard the P-3B by the University of Hawaii (Clarke et al.) using TSI 3-λ nephelometers corrected according to Anderson and Ogren (1998).Light absorption coefficients were measured aboard the NASA aircraft using 3-λ Radiance Research particle soot absorption photometers (PSAP's).The PSAP absorption measurements have been corrected using an updated algorithm (Virkkula, 2010), however levels of instrument noise remain ∼0.5 Mm −1 for a 240-300 s sample average, comparable to values reported previously (Anderson et al., 2003;McNaughton et al., 2009).Submicrometer aerosol extinction and absorption were measured on board the WP-3D by a team from the NOAA ESRL (Brock et al., and Lack) using a photo-acoustic method (Slowik et al., 2007) and a 532 nm Radiance Research PSAP also corrected according to Virkkula et al. (2010).

Results
The NASA DC-8 and P-3B sampled airmasses over the Western Arctic between 31 March and 19 April 2008 and include three sorties from Alaska to Greenland with one DC-8 flight via the North Pole (Fig. 1).The NOAA WP-3D flights occurred between 11 and 24 April 2008, and concentrated on sampling airmasses over Alaska and the nearby Arctic Ocean.All three platforms encountered airmasses containing anthropogenic urban/industrial emissions, biomass burning and mineral dust but with differing frequencies and intensities.
Mean vertical profiles of CO, dry aerosol extinction and accumulation mode (∼0.09-0.90µm) black carbon mass for each of the three airborne platforms are summarized in Fig. 2. Surface-based measurements of CO and total and submicrometer dry aerosol extinction at NOAA's Earth System Research Laboratory (ESRL) Barrow, Alaska site (Barrow) for April 2008 (Sharma et al., 2006), as well black carbon measurements at Environment Canada's Alert Weather Station (Alert) (Gong et al., 2010)  WP-3D are elevated aloft as a result of sampling intense biomass burning plumes over Alaska in late April (Warneke et al., 2009).Coefficients of variation for CO from the surface to 7 km are ∼10-25 % and illustrate the effect that intense episodic plumes have on the composition of the middle and lower troposphere.
The profiles of total and submicrometer dry aerosol extinction peak in the middle troposphere and illustrate the ubiquitous presence of mineral dust in the troposphere of the Western Arctic.Measurements of submicrometer extinction aboard the NOAA WP-3D illustrate the substantial variability in campaign-averaged extinction.Again, coefficients of variation generally increase from values ∼50 % near the surface and peak at 3-6 km where they can reach values >200 %.These differences result from spatial and temporal heterogeneity of aerosol found in pollution and biomass burning plumes, as well as differences in the sampling strategies employed by each aircraft.The differences cannot be attributed to differences in inter-platform instrument accuracy nor inlet performance as 4 pairs of intercomparison flights over a range of altitudes indicate instrument differences of generally less than 10 %.
The important role of combustion-derived aerosol is illustrated by the vertical profiles of BC mass in Fig. 2 by a smaller number of relatively intense plumes.For example, black carbon concentrations in the middle troposphere range from as low as 1-5 ng s m −3 to as much as 1500-1800 ng s m −3 resulting in coefficients of variation that are >100 %.The aircraft near surface (<2 km) concentrations of BC are lower but of the same magnitude as surface based measurements at Alert, Canada from 1997-2008.Thus while surface based observations provide time series needed to document trends in emissions and their transport to the Arctic, these measurements cannot, in general, be extrapolated to derive column burdens of BC.
Acetonitrile (ACN) (de Gouw et al., 2004;Warneke et al., 2006) and hydrogen cyanide (HCN) (Crounse et al., 2009) are commonly used to separate biomass burning (BB) influenced airmasses from airmasses dominated by urban/industrial emissions.The upper row of Fig. 3 plots black carbon mass versus CO mixing ratios colour coded by acetonitrile (ppbv).A priori we expect the relationship between BC and CO to vary as a function of their source (e.g.urban/industrial emissions versus biomass burning) and the airmasses transport history, i.e. dry versus wet convection followed by advection and dispersion.Relatively efficient combustion from urban/industrial sources are expected to have BC:CO slopes higher than those for relatively inefficient biomass burning sources.Similarly, while CO is approximately conserved during long-range transport, wet and dry deposition of aerosol during transport will result in lower BC:CO ratios for all plume types.As expected, the data from each aircraft indicate several distinct relations between BC and CO.While the DC-8 data have several small groups of data with high ACN, the presence of high ACN from BB plumes is more frequent in the WP-3D data.
The ratio of AMS accumulation mode measurements of organic aerosol to sulfate aerosol are also useful for determining whether an airmass is strongly influenced by biomass burning (Middlebrook et al., 2008).The bottom row of Fig. 3 replots BC versus CO with datapoints now colour coded by the base-10 logarithm of the ratio of AMS Organics:Sulfate.The use of the base-10 logarithm of the ratio is used in order to better illustrate the wide range of variability in the Organics:Sulfate ratio between 1:10 to as much as 20:1.The AMS data confirm that most airmasses sampled by the DC-8 and the P3-B are relatively low in organics while there are numerous airmasses from the WP-3D data with organics concentrations 3-20 times their sulfate content (log 10 = 0.5-1.3 in the plot).More detail regarding these biomass burning aerosol can be found in ARCPAC related publications including (Spackman et al., 2010;Warneke et al., 2009Warneke et al., , 2010)).
In an effort to separate Western Arctic airmasses dominated by biomass burning emissions from those dominated by urban/industrial pollution we select a threshold value of 160 pptv of acetonitrile and an AMS Organics:Sulfate ratio  of >2:1 (log 10 (Org:SO 4 ) >0.30) as two independent methods of stratifying the CO and black carbon data.We recognize that these values were chosen arbitrarily and that mixing and other processes will confound a true separation of airmass types.For example, the acetonitrile threshold is an absolute concentration and will subsequently be affected by airmass dilution, while the AMS-derived criteria is relative (a ratio), and will not change until plume concentrations approach the AMS instrument's limits-of-detection.However, the objective here is to compare/contrast the observations from each platform in an effort to describe, in general, the presence of absorbing aerosol in the Western Arctic and to what extent is may be derived from urban/industrial versus biomass burning sources.
Figure 4 plots histograms for all CO data acquired north of 55 • N for each platform.A subset of the histogram corresponding to the BB dominated plumes is over-plotted in red.The upper row of Fig. 4 uses Method-1, stratification using ACN, while the lower row uses Method-2, stratification based upon AMS ratios of Organics:Sulfate.Table 1 summarizes, as a percentage, the number of 60-s average data points classified as "biomass burning dominated" from the available CO data (N, at 60-s).The location and number of 60-s averaged data points recorded by the SP2 differs slightly from those recorded by the CO instruments.Figure 5 plots histograms of the base-10 logarithm of black carbon concentrations (µg s m −3 ) measured by the SP2 instruments, includ-ing the subsets of the data classified as BB according to each method of stratification.Table 2 summarizes as a percentage the black carbon data classified as BB from the available SP2 data for each aircraft.
Despite sampling over Alaska during a similar time period, the frequency with which each aircraft sampled biomass burning influenced airmasses is very different.This difference holds regardless of whether we base the assessment on the available CO or BC data or whether we stratify biomass burning dominated cases using acetonitrile or the ratio of AMS Organics:Sulfate.As evident in Tables 1 and 2, by using acetonitrile or AMS chemistry to stratify the data, we conclude that 11-14 % of the samples were dominated by BB emissions.Of greater interest is the implied proportion of BC over the Western Arctic which can be attributed to BB emissions versus BC produced from urban/industrial activities.These statistics are summarized in Table 3 where the observations from all three airborne platforms are combined.The BB-influenced airmasses, as defined here, account for ∼42-47 % of the total BC burden.
In-situ observations provide relatively accurate and precise measurements of absorbing aerosol along the aircraft flightpaths.However the limited duration (∼1-month) of the intensive observation period, as well as spatial and temporal variability of aerosol plumes means that aircraft data alone cannot be used to determine the representativeness of the sampling strategy, nor the seasonal or interannual variability   of absorbing aerosol over the Western Arctic.This consideration is particularly important as the ARCTAS/ARCPAC results demonstrate that Spring 2008 was a relatively intense year for BB emissions from Asia (Fisher et al., 2010;Warneke et al., 2009), with potentially reduced efficiency in the transport of urban/industrial emissions to the region (Fuelberg et al., 2010).Investigations into the spatial and temporal variability of BC over the Arctic domain are best suited to regional and global chemistry/aerosol models constrained by both   Inter-model diversity when simulating the atmospheric burden (load) of aerosol species at the poles is slightly lower for BC than for organic aerosol and dust (Textor et al., 2006).However, simulating organic aerosol is particularly complex (Kanakidou et al., 2005), with high diversity among models predicting their optical properties globally (Kinne et al., 2006).Given the important role of the crysophere in global climate and sea level, the in-situ airborne measurements collected by NASA and NOAA during ARCTAS and ARCPAC provide valuable constraints for future simulations of absorbing aerosol in the atmospheric of the Western Arctic.How-ever, due to its remote location, harsh environment, and atmospheric conditions that preclude comprehensive satellite coverage, the atmosphere of the Western Arctic likely remains critically undersampled.

Mass absorption efficiency of light absorbing carbon
Global climate models need to accurately simulate sources, sinks and the vertical distribution of absorbing aerosols in order to evaluate their radiative effects.Models also need to ensure that intensive aerosol parameters, e.g.single scattering albedo, are accurately simulated or parameterized (Anderson et al., 1999;Haywood and Shine, 1995).A key parameter linking BC emissions to their light absorbing properties is the mass absorption efficiency, with units of m 2 g −1 (MAE).To first order this quantity can be obtained from regressions of light absorption at a given wavelength against light absorbing mass.However, there are several factors which complicate this conceptually simple analysis for black carbon.First, although black carbon (BC) typically dominates aerosol absorption, absorption by organic carbon, e.g.humic-like substances (Graber and Rudich, 2006), and mineral dust (Lafon et al., 2006;Sokolik and Toon, 1996) can also contribute to total aerosol absorption.Second, absorption by internally mixed BC is enhanced by absorbing and non-absorbing coatings (Bond et al., 2006;Fuller et al., 1999;Lack and Cappa, 2010;Schnaiter et al., 2006).These coatings are common in both pollution (de Gouw and Jimenez, 2009;DeCarlo et al., 2010) and biomass burning plumes (Clarke et al., 2007;Gyawali et al., 2009).Finally, the accuracy of filter-based light absorption measurements (Cappa et al., 2008;Lack et al., 2008;Subramanian et al., 2007), the definitions of elemental versus black carbon, and the inter-comparability of their independent measurement are subjects of ongoing study (Schauer et al., 2003).
Despite these complications, regressions of total light absorption versus accumulation mode black carbon for each of 3-wavelengths (λ 470, 530, 660 nm) are relatively consistent between the NASA DC-8 and P-3B during ARCTAS (Fig. 6).MAE 470 shows the greatest disparity with a P-3B value of 14.8 m 2 g −1 , some 17 % higher than the 12.7 m 2 g −1 estimated from the DC-8 data set (black dashed line).DC-8 and P-3B MAE's at 530 are in good agreement with values of 9.5±0.2 and 10.3±0.4 m 2 g −1 respectively; values at 660 nm are also in good agreement at 6.7±0.1 and 6.9±0.3.While instrument uncertainty could potentially account for differences in blue absorption of 10-15 %, these effects are typically manifest at low concentrations due to instrument noise.The cause for the discrepancy is an erroneously low blue absorption value for the DC-8 PSAP instrument -a discrepancy identified during wingtip-to-wingtip intercomparisons among the aircraft.After empirically correcting the DC-8 blue absorption data (using NASA P-3B data) the slope and uncertainty of the MAE 470 regression are recalculated at 15.2±0.3m 2 g −1 .The corrected data are used throughout the remainder of this manuscript.
As noted above, these MAE estimates represent total absorption efficiency and include absorption by non-BC carbonaceous species, mineral dust, and possible amplification due to coatings.In an effort to better constrain and separate the MAE's for BC and brown carbon (BrC), we take a closer look at the NASA P-3B data.First, the 1-min averages are pooled into 2-min averages and a lower threshold of 3 Mm −1 for total scattering (550 nm) is used to exclude low signal-tonoise absorption data (typically PSAP noise is ∼0.5 Mm −1 for a 300 s sample).Next the ratio of submicrometer to total scattering is used to reduce the influence of high dust cases, i.e. we discard samples with a fine-mode fraction of scattering (FMF scat ) below 0.6 (Anderson et al., 2003;Mc-Naughton et al., 2009).After applying this stratification the sample number is reduced by about half to N = ∼700.
The first column of Fig. 7 replots the P-3B total MAE for light absorbing carbon (LAC) using these criteria.The top row of data are color-coded by absorption Ångström exponent (steep dependence = red; shallow dependence = blue), the center row is colored by supermicrometer mass fraction (coarse-mode dominated = red; fine-mode dominated = blue), and the final row is colored by scattering Ångström exponent (coarse-mode dominated = red; finemode dominated = blue).After performing this stratification we find that the MAE's for total aerosol change very little compared to the bulk analysis in Fig. 6 and therefore conclude that total absorption is dominated by light absorbing carbon.
The wavelength dependence of black carbon is relatively weak and expected to vary as λ −1 (Kirchstetter et al., 2004).Absorption by brown carbon is typically very small at 660 nm where we argue (Clarke et al., 2007) that it is primarily due to black carbon.Making use of this assumption and the PSAP measurements of absorption at 660 nm, it is possible to compute excess light absorption at 470 and 530 nm, i.e. absorption which is potentially due to brown carbon and having a wavelength dependence in excess of λ −1 .We can also use the HiGEAR thermal volatility measurements of the aerosol size distribution to determine submicrometer refractory volume (420 • C, τ ∼ = 0.1 s) (Clarke et al., 2007(Clarke et al., , 2004;;McNaughton et al., 2009).The refractory volume is then converted to an estimate of non-BC refractory mass (M ref ) by assuming it has a density of 1.3 g cm −1 and then subtracting the mass of BC measured concurrently by the SP2.
The second panel of Fig. 7 plots excess absorption versus M ref and the regressions based upon them.Using this technique we compute an MAE for the refractory mass of 0.83±0.02and 0.27±0.01m 2 g −1 for 470 nm and 530 nm respectively.Here±values represent uncertainties to the regression.When instrument noise, flow rate uncertainties etc. are added in quadrature, the uncertainty estimates are 0.83±0.15and 0.27±0.08m 2 g −1 for 470 nm and 530 nm wavelengths.As the computations are normalized to the absorption measurement at the 660 nm wavelength, we cannot compute an MAE value at 660 nm.
The actual constituents of M ref will include both refractory organic matter as well as non-organic submicrometer refractory aerosol from urban/industrial and natural sources.These constituents may or may not be responsible for enhanced absorption at short wavelengths.However, the data were pre-screened to reduce the influence of absorbing dust by only using data for which FMF scat >0.6.We also recognize that some light-absorbing organic species may volatilize below our size distribution pre-heater temperature of 420 • C (Andreae and Gelencser, 2006).However, in the absence of more detailed chemical information, we are explicitly assuming that M ref ∼ =brown carbon (BrC).
To evaluate the validity of this assumption, the MAE values calculated using the ARCTAS data can be compared to MAE values for brown carbon from INTEX-A (Clarke et al., 2007) as well as those computed by Yang et al. (2009) during EAST-AIRE (Li et al., 2007).Note that the Yang et al. absorbing mass values are based on EC/OC measurements using the Sunset Labs technique (Huebert et al., 2004a;Kline et al., 2004), while their absorption measurements at 470, 520 and 660 nm use a corrected 7-wavelength aethalometer (Arnott et al., 2005;Bond et al., 1999).The Yang et al., aethalometer data were compared to a corrected (Virkkula et al., 2005) 3-λ PSAP and showed excellent agreement (Yang, 2007).However, the updated PSAP correction (Virkkula, 2010) means that we can expect both the INTEX-A and the EAST-AIRE estimates of MAE's to be low by ∼10-30 %.Also note that the Yang et al.MAE BrC values at all wavelengths (e.g.0.32 m 2 g −1 @ 660 nm) are normalized to BC absorption at 990 nm rather than 660 nm.Thus to fairly compare these values we need to adjust the ARCTAS brown carbon MAE values by adding 0.32 m 2 g −1 to the ARCTAS MAE values at 470 and 530 nm's.Estimates of absorption by this refractory organic matter, i.e.BrC, are summarized in Table 4 and show good agreement between independent measurements using different techniques, at different locations, and during different seasons.
By measuring the refractory mass and establishing an MAE for brown carbon, we can estimate the absorption by brown carbon for each sample.We then subtract this value from total aerosol absorption in order to derive a better estimate for the BC-only absorption.Figure 8 summarizes this analysis by regressing total absorption, minus absorption from brown carbon, against SP2 measurements of BC.Once again the upper, middle and lower panels are colour coded according to absorption Ångström exponent (top), coarse mode scattering fraction (middle) and scattering Ångström exponent (bottom).Regressions remain robust, even improving slightly.The success of this method for refining the estimated MAE BC is evident for the cluster of highest absorption Ångström exponent (∼2.5) data in the upper left panel of Fig. 7. Absorption of blue light (470 nm) by this cluster of data differs significantly from the initial regression.By subtracting the apparent absorption by brown carbon (product of brown carbon MAE and mass of M ref ) this cluster of data moves closer to the regression and improves the R 2 by ∼3 %.Taking the Ångström exponent of the regression slopes we can verify that the regression is approximately preserving a 1/λ dependence (i.e.−log(11.2/7.4)/log(470/660= 1.2).PSAP and SP2 flow uncertainties (<5 %) and uncertainties of the regression (+/-0.2 to 0.3 m 2 g −1 ) are smaller than differences in absorption measured during wingtip-towingtip aircraft intercomparisons (−8 % to +13 %).We also performed a sensitivity analysis given an uncertainty for the wavelength dependence of absorption by black carbon of 1/λ (0.9−1.6) .If we assume that these uncertainties add in quadrature, then our best estimates for ARCTAS MAE BC for each wavelength are 11.2±0.8,9.5±0.6 and 7.4±0.7 m 2 g −1 .
We can now compare the ARCTAS MAE values for BC to those already reported in peer-reviewed literature.
Our calculations for the MAE of BC account for filter matrix effects (Virkkula, 2010;Virkkula et al., 2005), but do not account for absorption enhancements due to coatings.Thus results report here should be viewed as the mass absorption efficiency of dry (RH<40 %) aerosol found in the atmospheric environment rather than an MAE for uncoated graphitic carbon or values for pure black carbon particulate calculated from theory.Coating of BC particulate is expected to result in ∼35 % absorption enhancement depending on the ratio of shell to core diameters (Fuller et al., 1999), or as much as a factor of 1.8-2.0depending on the wavelength of interest and nature of the coatings (Lack and Cappa, 2010;Schnaiter et al., 2005).
If we assume that coating by non-or weakly-absorbing aerosol results in a ∼35 % enhancement then our reduced MAE value of 7.0 +/ 0.6 m 2 g −1 at 530 nm is in good agreement with the value of 7.5±0.3m 2 g −1 (@550 nm) reported for BC in the review by Bond et al. (2006) and 7.1 m 2 g −1 recently reported for diesel soot (Adler et al., 2010).If absorption by uncoated BC is being over estimated by 80 % then a reduced value of ∼5.3 m 2 g −1 is comparable to Fuller et al. (1999) who reported an MAE value of 5.4 m 2 g −1 for amorphous graphitic spheres from diesel soot.The ARC-TAS/ARCPAC value of ∼5.3 m 2 g −1 is also effectively identical to Kondo et al. (2009) who reported 5.7±0.3m 2 g −1 and recently measured values of 5.5 m 2 g −1 (Kondo et al., 2011).A more detailed analysis of potential absorption enhancements using the aerosol size distributions and SP2 measurements of the BC size distributions during the ARC-TAS/ARCPAC field campaigns is warranted, and should form the basis of future analyses.

Mineral Dust over the Western Arctic
A salient observation from the ARCTAS/ARCPAC field campaigns is the ubiquity of supermicrometer aerosol in the free troposphere (FT) of the Western Arctic.The first panel of Fig. 9 plots vertical profiles of NASA P-3B (red) and the NOAA WP3D (blue) supermicrometer volume (V coa ).Mean and median volume concentrations are likely higher aboard the NOAA WP-3D because of the use of an active low-turbulence inlet (Huebert et al., 2004b;Wilson et al., 2004) with a 50 % passing efficiency larger than the passive solid diffuser inlets used aboard the NASA DC-8 and P-3B (McNaughton et al., 2007).Intense plumes of coarse mode aerosol were rare during ARCTAS, however the WP-3D did encounter one such plume with 60-s average volume greater than 50 µm 3 m −3 .The effect of this plume on the mean value of V coa for the WP-3D is shown in Fig. 9 compared to the mean P-3B data.
The second panel of Fig. 9 summarizes the mean fine mode fraction (FMF) of extinction and aerosol volume North of 55 • N during ARCTAS/ARCPAC.Measurements of total and submicrometer extinction are common to the NASA aircraft only, and are plotted as solid green (DC-8) and blue (P-3B) lines; supermicrometer volume is common between the two P-3's and plotted as dashed lines.The DC-8 spent the most time at high altitude and in the High Arctic (i.e.North of 70 • N), whereas the two P-3's spent more time in the mid-and lower-troposphere and concentrated their sampling near Alaska.The DC-8 submicrometer nephelometer also periodically malfunctioned during the campaign, resulting in fewer samples in Fig. 9  that supermicrometer aerosol concentrations are highest between 6 and 8 km, decreases toward the surface but then increases again below ∼1 km.The PALMS instrument aboard the WP-3D measured supermicrometer particle chemistry during ARCPAC, and provides the unique ability to separate these aerosol into dust and sea salt particle types (Froyd et al., 2009).Above ∼4 km mineral dust particles account for >80 % of the total number of particles analyzed by the PALMS.Below 4 km the number fraction of dust particles then decreases, such that sea salt particle types account for 60-80 % of the particle number below 1 km.These results can be put further into context by examining the vertical profiles of fine mode fraction of scattering (FMF scat ; Anderson et al., 2003) measured during other airborne field campaigns, employing the same inlets and/or instrumentation, while sampling over the North Pacific and North America in the Spring and Summer (Clarke and Kapustin, 2010).Figure 10 plots FMF scat profiles measured during ACE-Asia (Clarke et al., 2004;Huebert et al., 2003), INTEX-NA (Clarke et al., 2007;Shinozuka et al., 2007), INTEX-B/IMPEX (Dunlea et al., 2009;McNaughton et al., 2009), and ARCTAS.In the figure the analysis is restricted to cases where total light scattering is greater than 3 Mm −1 to eliminate ratios derived from low signal-to-noise scattering values which can skew the summary statistics.We plot the fractions with respect to atmospheric pressure to better normalize measurements occuring over different terrain and during different seasons.
The ACE-Asia measurements (Spring, 2001) of the East Asian source region are dominated by "mixed airmasses" (0.3<FMF scat <0.6) until the 700 mb level where the average FMF scat drops below 0.5 and "dusty airmasses" (FMF scat <0.3) are more common.Measurements aboard the NASA DC-8 over the North Pacific during INTEX-B (Spring 2006) and aboard the NSF/NCAR C-130 off the US west-coast during IMPEX (Spring, 2006) are qualitatively very similar.Both are influenced by sea salt below 900 mb, show "fine-mode dominated" (FMF scat >0.6) or mixed airmass types up to 700 mb and then become more heavily influenced by mixed airmass types above 700 mb due to the long-range transport of Asian pollution and dust.In contrast measurements over North America east of 100 • W longitude during INTEX-NA (Summer, 2004) show a very low frequency of airmasses that contain supermicrometer dust.Those INTEX-NA samples that did contain dust (rather than sea salt near the surface) were either fresh urban plumes, or located in the US Southwest where airmasses contained traces of Saharan rather than Asian dust.During ARC-TAS the NASA DC-8 spent the most time in the High Arctic whereas the two P-3B's spent most time sampling near Alaska.The ARCTAS DC-8 FMF scat profile also suffers from low sample numbers due to malfunction of the submicrometer nephelometer.However, the data still indicate a greater proprtion of "mixed" airmass types were found above 700 mb.The ARCTAS P-3B data indicate that fine mode dominated and mixed airmass types are comparable in number below ∼700 mb but that mixed and even dusty airmass types become more prevalent in the middle to upper free troposphere.Thus, the ARCTAS/ARCPAC data support the hypothesis that Western Arctic airmasses are heavily influenced by emissions of Asian Dust transported across the Pacific after lofting by mid-latitude cyclones over Central Asia, East Asia and the Eastern North Pacific (Fuelberg et al., 2010;Stohl, 2006).
In an attempt to derive the MAE for supermicrometer dust, an analysis similar to that conducted for light absorbing carbon was undertaken.Despite the ubiquity of supermicrometer aerosol over the Western Arctic, the low absorption efficiency of dust and the high noise levels of the PSAP's make an accurate determination of the dust MAE difficult. of data result in MAE estimates of 0.034 m 2 g −1 at 470 nm, and 0.017 m 2 g −1 at 530 nm.Even after heavy stratification the regressions statistics remain poor (R 2 = 0.29, 0.13) but the slopes differ significantly from zero with 95 % confidence intervals for the slopes of 0.023-0.046m 2 g −1 , and 0.008-0.027m 2 g −1 respectively.Yang et al. (2009), report an MAE 470 for dust of 0.050 m 2 g −1 , greater than our highly uncertain estimate of 0.034 m 2 g −1 and slightly greater than our 95 % confidence interval.Reported MAE 530 values for Asian dust range between 0.009 and 0.034 m 2 g −1 (Clarke et al., 2004;Yang et al., 2009).These are plotted as dashed (C04) and solid (Y09) black lines in Fig. 11 and roughly bracket the 95 % CI of our estimate.

Relative contributions of sources and absorbing species to light absorption over the Western Arctic
The efficacy of radiative forcing by BC, BrC and mineral dust over Polar Regions depends on their vertical distribution in the atmosphere.Although BC is the dominant absorber, its vertical distribution differs from that of dust and from non-absorbing sea salt.Furthermore, the strong wavelength dependence of both brown carbon and dust mean that the proportion of light absorption due to these species will also vary as a function of solar wavelength.Thus it is useful to compare the vertical distribution of the relative contribution from each source and species relative to absorption from all sources.
The first panel of Fig. 12 summarize the mean and median mass concentration profiles for BC over the Western Arctic.The ARCTAS/ARCPAC measurements show that black carbon peaks in the middle troposphere between ∼2 and 6 km www.atmos-chem-phys.net/11/7561/2011/Atmos.Chem.Phys., 11, 7561-7582, 2011  with values in excess of 0.1 µg s m −3 .The second panel summarizes the vertical distribution of supermicrometer mass after classifying these aerosol as either dust or sea salt.This classification is accomplished by using data from the PALMS instrument (Froyd et al., 2009) to separate the P-3B and WP-3D supermicrometer number distributions into aerosol types, and then applying realistic assumptions for the shape of lognormal dust and sea salt number distributions over the North Pacific (number median diameter, NMD DU = 0.65 µm, geometric standard deviation, σ g,DU = 2.0; NMD SS = 0.40 µm, σ g,SS = 2.15) (McNaughton, 2008).These number distributions are then converted to volume distributions, the integrals multiplied by appropriate dry bulk densities (ρ DU = 2.06 and ρ SS = 2.20 g cm −3 ), and the statistics summarized to construct the vertical profiles in Fig. 12.As indicated, mineral dust concentrations greater than 2 µg cm −3 are common above 4 km and decrease linearly to less than 0.5 µg sm −3 at the surface.Sea salt is most common below ∼1 km but was detected throughout the FT by the PALMS instrument.Table 5 contains summary statistics for the vertical profiles of aerosol mass measured during ARCTAS/ARCPAC including mean, median and standard deviations of BC mass, submicrometer mass, submicrometer refractory mass (420 • C, τ = 0.1 s), and supermicrometer mineral dust and sea salt mass.An estimate of BrC mass can be obtained by subtracting BC mass from submicrometer refractory mass.
The third panel of Fig. 12 plots vertical profiles of aerosol single scattering albedo at 470, 530 and 660 nm wavelengths.Low scattering Ångström exponent but high absorbing Ångström exponent by Asian dust at high altitude leads to lower 470 nm than 660 nm SSA's between 6 and 8 km.In the lower free troposphere (2-6 km), dust mass is decreasing as black carbon concentrations, especially from BB emissions, are reaching their peak values.Here the SSA's are equal at each wavelength.Near the surface, we observe the lowest masses of LAC (mostly urban/industrial emissions) and enhanced scattering due to sea salt aerosol.The result is higher dry SSA's that increase at shorter wavelengths, an effect that will be magnified in the ambient atmospheric environment due to high RH in/near the boundary layer.Table 6 contains summary statistics for dry aerosol optical properties measured during ARCTAS including total extinction, column aerosol optical depth (AOD), single scattering albedo and extinction-weight column SSA values at 470, 530 and 660 nm.ARCPAC extinction values are not included as they are submicrometer measurements only.However, a comparison between the NASA DC-8 plus P-3B submicrometer extinction values and submicrometer extinction for all three aircraft (i.e.including the NOAA WP-3D values) is of interest because the NOAA WP-3D sampled a much higher proportion of BB influenced airmasses (Tables 1-3).Submicrometer AOD 530 values for the NASA aircraft average 0.12 (median = 0.08) with an extinction weighted column SSA 530 value of 0.94±0.5.The submicrometer AOD 530 values computed from the average of all three aircraft average 0.16 (median = 0.10) with an SSA 530 value of 0.95±0.04.These values can be compared to the total AOD 530 and SSA 530 values from the NASA aircraft which averaged 0.15 (median = 0.10), and 0.95±0.3.Thus, the presence of the intense but episodic BB plumes observed during ARCTAS/ARCPAC has increased the column AOD values by ∼1/3 while potentially increasing the column SSA values by 0.01; the result of typically higher SSA for BB aerosol at 530 nm compared to 530 nm SS for urban/industrial emissions.
There are too few dust-free samples to calculate a robust vertical distribution of submicrometer non-BC refractory mass (i.e.BrC).To calculate this quantity we instead use the average ratio of BC to submicrometer refractory mass (BC/(BrC+BC)) to estimate BrC mass, restricting our analysis to cases where FMF scat >0.6.In urban/industrial airmasses the ratio of black carbon to submicrometer nondust refractory mass is approximately 0.25 (0.27 in Clarke et al., 2007;0.23 during ARCTAS).In biomass burning plumes this ratio is closer to 0.15 (0.13 in (Clarke et al., 2007;0.17 during ARCTAS).We can then use mass profiles of BC and dust to separate the contributions of each absorbing species at 470 and 530 nm wavelengths.Based on ARCTAS/ARCPAC results at 470 nm we assume an MAE BC of 11.2 m 2 g −1 , an MAE BrC value of 0.83 m 2 g −1 , and an MAE Dust of 0.034 m 2 g −1 .At 530 nm we use MAE values of 9.5, 0.27 and 0.017 m 2 g −1 for each species.
The solid blue and green lines in Fig. 13 plot the relative contribution of absorption by LAC to total absorption at 470 nm (left) and 530 nm (right) wavelengths.The figure further separates the contributions from the two main sources of anthropogenic absorbing aerosol -urban/industrial (red lines), biomass burning (black lines) -and breaks the LAC contribution into the fraction of total absorption from black carbon (solid red, black) and brown carbon (dashed red, black) at both wavelengths.
Above 6 km the contribution of mineral dust to total light absorption is significant, and constitutes 4-12 % or 3-8 % of absorption at 470 and 530 nm wavelengths respectively.At these altitudes, light absorption by BrC from urban/industrial emissions average 9 % (470 nm) and 4 % (530 nm) of total light absorption, whereas absorption by BrC from biomass burning emissions accounts for 12 % and 6 % of total light absorption at each wavelength.Thus when combined, light absorption by BrC from anthropogenic sources is responsible for a higher proportion of total light absorption than nominally "natural" sources of mineral dust.Below 6 km the contribution by dust to total light absorption is even lower.LAC from urban/industrial sources constitute 51-54 % of total light absorption, while LAC from BB emissions account for 42-45 % of light absorption by aerosol.Thus the 2008 ARCTAS/ARCPAC results show that largely anthropogenic sources of LAC account for 94-99 % of total light absorption at 470 nm, and 96-99 % of total light absorption at 530 nm in the atmosphere below 6 km over the Western Arctic.
Future model simulations should seek to replicate the ARCTAS/ARCPAC vertical profiles of aerosol optical properties, specifically the contributions of BC, BrC, mineral dust and sea salt to total extinction and single scattering albedo, in order to determine the effects of these man-made absorbing aerosol on radiative budget of the entire Arctic domain compared to radiative forcing by dust alone in an unperturbed pre-industrial Arctic atmosphere.

Conclusions
In April of 2008 NASA's ARCTAS field campaign and NOAA's ARCPAC field campaign deployed research aircraft to the Western Arctic as part of POLARCAT, a core International Polar Year activity.Despite sampling similar geographic regions over a comparable period of time, a great deal of variability in combustion tracers (e.g.CO, light extinction, black carbon mass) was observed between the aircraft.The major cause of this variability was early, extensive agricultural and forest fires in Kazakhstan andSiberia in April 2008 (Fisher et al., 2010;Warneke et al., 2010).These plumes were sampled repeatedly by the NOAA WP-3D after 19 April, but sampled only briefly by the NASA DC-8 and P-3B on 19 April.
After pooling the results from all three aircraft, we conclude that total light absorption over the Arctic was predominantly (88-99 %) due to light absorbing carbonaceous species derived from urban/industrial activities and biomass burning.Using the gas phase tracer acetonitrile and measurements of submicrometer aerosol chemistry, we were able to determine that airmasses dominated by biomass burning aerosols account for just 11-14 % of the air volume sampled.However, these inefficient open combustion sources account for 42-47 % of the total burden of black carbon over the Western Arctic, and were responsible for 39-44 % of total light absorption; total light absorption being the sum of absorption by black carbon, brown carbon and mineral dust.Mineral dust was found to be ubiquitous throughout the troposphere of the Western Arctic.Mineral dust concentrations below 2 km averaged less than 0.5 µg s m −3 increasing to values of 1.0 to 2.8 µg s m −3 from 2 to 6 km.However, below 6 km light absorption by carbonaceous species dominates the total absorption term so that dust accounts for only 1-3 % of total light absorption at these altitudes.Above 6 km dust concentrations reach average values as high as 2.1-3.4 µg s m −3 .And though concentrations of LAC are decreasing above 6 km, total light absorption remains dominated by LAC such that absorption by mineral dust accounts for only 4-12 % or 3-9 % of total absorption at 470 nm or 530 nm wavelengths respectively.Light absorption by brown carbon is a higher fraction of total absorption for BB plumes, accounting for ∼12 % or ∼6 % of total light absorption at 470 and 530 nm wavelengths.The corresponding contributions of total absorption by BrC derived from urban/industrial sources are ∼10 % at 470 nm and ∼4 % at 530 nm.
Comparisons between intensive and extensive aerosol properties among the aircraft are excellent.Stratification of the data allowed us to derive mass absorption efficiencies for black carbon and brown carbon as a function of wavelength.At 470, 530 and 660 nm the MAE BC values are 11.2±0.8,9.5±0.6 and 7.4±0.7 m 2 g −1 .These estimates represent insitu values for dry atmospheric aerosol and are consistent with recent estimates of 35-80 % enhancements in 530 nm absorption due to coating of pure black carbon particulate with clear or weakly absorbing coatings.Assuming that refractory mass measured in relatively dust free samples is dominated by refractory organic matter responsible for enhanced absorption at shorter wavelengths, we were able to estimate MAE values for brown carbon at 470 and 530 nm wavelengths of 0.83±0.15and 0.27±0.08m 2 g −1 .These brown carbon MAE values are broadly consistent with recently reported data from the East-Asian source region.An attempt to derive the mass absorption efficiencies for Asian dust produced values of 0.034 m 2 g −1 and 0.017 m 2 g −1 .However, these estimates are highly uncertain (95 % CI of 0.023-0.046and 0.008-0.027m 2 g −1 , with R 2 <0.29, 0.13) due to the limitations imposed by the filter-based PSAP instrument.
Recent modeling studies show black carbon is poorly simulated in the Arctic compared to its simulation at midlatitudes (Koch et al., 2009), that simulating organic aerosol is particularly complex (Kanakidou et al., 2005), and that there is high diversity among models predicting their optical properties globally (Kinne et al., 2006).The ARC-TAS/ARCPAC results show that while total light absorption is dominated by black carbon, predominantly anthropogenic sources of brown carbon contribute a larger proportion of total light absorption than nominally natural sources of mineral dust.Above 6 km the proportion of total absorption by brown carbon at 470 nm wavelength can exceed that of mineral dust by factors of 2 to 5 even though mineral dust transported long range from East Asia has its highest concentrations at these altitudes.Below 6 km dust concentrations decrease while LAC concentration increase such that absorption by BrC can be 10-15 times the absorption by dust and absorption by total LAC can be 50-100 times that of 'natural' Asian dust.
The remote location, harsh environment, and prevailing meteorology over the short duration of ARCTAS/ARCPAC field campaigns, mean that the atmosphere of the Western Arctic likely remains critically undersampled.These factors preclude using the airborne data to evaluate seasonal or interannual variability -the role of models.Future investigations by climate modeling teams should seek to replicate the relative proportions of these absorbing species over the Western Arctic, if not their concentrations at these locations, in an effort to determine the effects of man-made emissions of absorbing aerosol on the radiation balance of the entire Arctic domain.
C. S. McNaughton et al.: Absorbing aerosol in the troposphere of the Western Arctic
. The mean BC profiles peak in the middle troposphere with concentrations between 200-300 ng s m −3 (T = 273.15K, P = 1013.25 mb), but are reduced by a factor of four to ∼50 ng s m −3 at the surface.Median values (not shown), are about half the mean values, indicating the data are skewed C. S. McNaughton et al.: Absorbing aerosol in the troposphere of the Western Arctic Campaign mean vertical profiles of (left) carbon monoxide (+/-1σ dashed), (middle) total and submicrometer (solid & dashed) dry aerosol extinction, and (right) mean black carbon as measured aboard the NASA DC-8 (green), P-3B (red) and NOAA WP-3D (blue) during ARCTAS/ARCPAC in April of 2008.

Fig. 4 .
Fig. 4. Histograms of CO over the Western Arctic during April 2008.Top: CO data stratified using a threshold concentration of CH 3 CN of 0.160 ppbv to identify BB airmasses.Bottom: CO data stratified using a threshold ratio of AMS Organics:Sulfate of 2.0 to identify BB airmasses.

Fig. 5 .
Fig. 5. Histograms of base-10 logarithm of BC concentrations over the Western Arctic during April 2008.(top-row) BC data stratified using a threshold concentration of CH 3 CN of 0.160 ppbv to identify BB airmasses.(bottom row) BC data stratified using a threshold ratio of AMS Organics:SO 2− 4 of 2.0 to identify BB airmasses.

Fig. 6 .
Fig.6.Apparent mass absorption efficiency of BC in Arctic aerosol using total aerosol absorption at each of the 3 PSAP wavelengths; DC-8 data on the left, P-3B data on the right.DC-8 data at 470 nm wavelength (black symbols, dashed line) have been corrected (blue symbols, solid line) based on wingtip-to-wing-tip intercomparisons with the P-3B.
Total mass absorption efficiency at 3-wavelengths for light absorbing carbon (left) and for brown carbon (right), normalized to absorption at 660 nm.Rows are color-coded by absorption Ångström exponent (top), coarse mode fraction (middle), and scattering Ångström exponent (bottom).

Fig. 9 .
Fig. 9. (left) Profiles of mean and median supermicrometer volume (V coa ) from the NASA P-3B and the NOAA WP-3D.(right) Profiles of fine-mode fraction of extinction (solid lines) and aerosol volume (dashed lines) -Note that N = 600 for the DC-8 extinction profile due to instrument malfunction.For comparison N = 2000 for the P-3B and N = 1500 for the WP-3D.

Fig. 10 .
Fig. 10.Vertical profiles of the fine mode fraction of scattering (FMF scat ) measured in-situ via aircraft during recent NASA and NSF-funded airborne field campaigns.Data points (N) are restricted to cases where total scattering is greater than 3 Mm −1 to eliminate ratios computed from low signal-to-noise scattering values.

Fig. 11 .
Fig. 11.Estimate for the MAE of supermicrometer dust at 470 nm (top) and 530 nm (bottom).Data were heavily stratified symbols) so as to include only samples that were dominated by dust.Solid blue and green lines are the regressions, dashed are the 95 % confidence intervals for the slope.Yang et al. (2007) (solid black) and Clarke et al. (2004) (dashed black), are included for comparison.

Fig. 13 .
Fig. 13.Relative contributions of light absorbing carbon to total absorption as a function of altitude over the Western Arctic at 470 nm (left) and 530 nm (right) wavelengths.Contributions by black carbon (solid lines) and brown carbon (dashed) are further separated by their sources, urban/industrial emissions (red), and emissions from biomass burning (black).

Table 1 .
Percentages of airmasses classified as biomass burning for each aircraft.Two methods of stratification are employed in order to characterize all of the available date for carbon monoxide (CO).There are no measurements of acetonitrile aboard the NASA P-3B.Platform Method-1 ACN >0.160 ppbv Method-2 Org:SO4 >2.0

Table 2 .
Percentages of airmasses classified as biomass burning for each aircraft.Two methods of stratification are employed in order to characterize all of the available date for black carbon mass (BC).There are no measurements of acetonitrile aboard the NASA P-3B.
in-situ measurements and satellite retrievals (e.g.Fisher et  al., 2010).Climate models which include aerosols and that demonstrate reasonable fidelity with respect to in-situ measurements obtained during the IPY could be used to assess the representativeness of the ARCTAS/ARCPAC aircraft sampling, potentially improving future collaborations.However recent comparisons between global models and insitu airborne measurements of BC highlight the huge diversity among model simulations, particularly over the Arctic domain(Koch et al., 2009).

Table 3 .
Burden, expressed as a percentage of total burden, of black carbon mass associated with biomass burning plumes compared to the mass of black carbon for all airmasses sampled during ARC-TAS/ARCPAC.

Table 4 .
Summary of mass absorption efficiencies for refractory organic material assumed to be BrC normalized to absorption at 660 nm and to absorption at 990 nm.By definition the ARCTAS 660 nm MAE is equal to the EAST-AIRE value of 0.32 m 2 g −1 when the data are normalized to BC absorption at 990 nm wavelengths.

Table 5 .
Summary of vertical profiles for black carbon mass, submicrometer aerosol mass, refractory submicrometer aerosol mass, as well as supermicrometer mineral dust and sea salt aerosol mass.Statistics include mean, median and standard deviations calculated from all available 1-min averaged data measured onboard the NASA DC-8, P-3B and NOAA WP-3D during ARCTAS/ARCPAC.

Table 6 .
Summary of vertical profiles for total light extinction and aerosol single scattering albedo at 470, 530 and 660 nm wavelengths.Statistics include mean, median and standard deviations from all available 1-min averaged data measured onboard the NASA DC-8, P-3B during ARCTAS.Column dry aerosol optical depth (AOD) and SSA values are weighted by extinction.